Few of life’s experiences evoke greater apprehension than a diagnosis of Alzheimer’s disease (AD). Virtually unknown to the public until the 1980s, it is alone among the 10 most common fatal diseases of developed nations in lacking a disease-modifying treatment. AD affects people of all ethnicities; in the United States, African Americans have twice the prevalence of European Americans (1). The cumulative financial cost to society of late-life dementias (of which AD comprises тИ╝60%) is estimated to exceed those of heart disease and cancer (2). This dismal reality may now be changing. The properties of the key proteins comprising the amyloid plaques [amyloid-╬▓ (A╬▓)] and neurofibrillary tangles (tau) that define the neuropathology of AD have been identified. Coupled with extensive genetic studies, a sequence of lesion formation in brain networks serving memory and cognition is suggested. Antibodies that target these proteins are in advanced trials, and aducamumab, which clears A╬▓, was recently approved, though not without controversy.
Through longitudinal analyses of humans with rare, causative mutations in APP (the A╬▓ precursor protein) and presenilin (the catalytic subunit of ╬│-secretase, which cleaves APP to generate A╬▓), it has become clear that biochemical alterations in the brain begin at least two decades before cognitive symptoms develop. During this long presymptomatic interval, extracellular accumulation of the self-aggregating A╬▓42 peptide into initially soluble oligomers and then increasingly large polymers and insoluble fibrils is accompanied by binding of the oligomers to the plasma membranes of microglia, astrocytes, and myriad neurites and synapses (see the figure). Although this amyloid hypothesis of AD is often drawn linearly for simplicity (3), many of the changes likely arise in temporal proximity (4).
Genome-wide association studies in typical late-onset AD (i.e., after age 65) have converged on risk alleles in diverse genes mediating cholesterol and lipid regulation, synaptic network functions, and especially microgliosis (altered microglia) and neuroinflammation. The most potent genetic risk factor is the apolipoprotein E (APOE) ╧╡4 variant: Heterozygosity raises AD risk 2- to 5-fold, and homozygosity increases it >5- to 10-fold. Its pathogenic mechanism appears to involve decreased glial-mediated clearance of A╬▓ from the brain’s extracellular space, leading to more amyloid in cerebral plaques and microvessels (5). In mice, the APOE4 protein can also promote tau-mediated neurodegeneration and glial activation, both in the presence and absence of amyloid (6). Some other AD genetic risk factors have likewise been linked to enhanced A╬▓ deposition and/or the macrophage and microglial reaction to it.
Two decades ago, theories about AD pathogenesis seemed divided over the primacy of amyloid versus tau deposition. This false dichotomy has been supplanted by a growing consensus that A╬▓ aggregation in the brain [indicated by declines in soluble A╬▓ monomers in cerebrospinal fluid (CSF) and accrual of insoluble plaques seen on amyloid-PET (positron emission tomography) scans] begins early in people destined to develop AD and is followed by glia-mediated inflammation and the accumulation and spread of tau tangles in brain regions that serve cognition (7, 8). Rising amounts of extracellular A╬▓ lead to aggregates, including soluble oligomers, that appear to enhance the accrual of tau tangles and altered neurites beyond the medial temporal lobe, where these lesions are often present in older people without AD. Such tau accumulation and spread in the brain, perhaps via neuron-to-neuron connections, seems necessary for the development of cognitive symptoms in AD (9). In APP transgenic mice, deletion of the gene that encodes tau does not alter amyloid plaques but significantly lessens their behavioral consequences. Thus, A╬▓ oligomerization appears to initiate AD neuropathology, leading to altered tau in neurites and cell bodies as well as microgliosis and blood monocyte infiltration into the brain.
The failure to reach primary and secondary outcomes in numerous trials of potentially AD-modifying agents may be explained in one or more ways: failure of the agent to achieve robust and selective target engagement in the brain; initiating treatment at a clinical stage that is too advanced to be effective; underpowered trials; adverse side effects on cognition; and faulty trial execution. The precise reasons differ among the unsuccessful trials to date. But a few recent trials appear to have met their primary endpoints or come close to them and have also achieved some secondary endpoints.
Alzheimer’s disease neuropathology includes extracellular amyloid plaques containing myriad amyloid-╬▓ (A╬▓) oligomers and intraneuronal tangles containing phosphorylated tau. Microglia and astrocytes become activated, leading to neuroinflammation and the spread of neuropathology. Antibodies to A╬▓, administered intravascularly, can clear amyloid plaques.
GRAPHIC: KELLIE HOLOSKI/SCIENCE
The clearest evidence of disease modification so far has come from secondary biomarker endpoints, principally a substantial decrease in amyloid plaques over 18 months, as measured by amyloid-PET. For example, this occurred in the two phase 3 trials of aducanumab, an A╬▓ monoclonal antibody that was approved by the US Food and Drug Administration (FDA) on 7 June 2021. Additional biomarker changes included a decrease in the elevated CSF concentration of phosphorylated tau protein and a reduction of brain tau-PET signal, but these outcomes were only measured in a small minority of aducanumab recipients. Although the marked decrease in amyloid deposits can be viewed as biological evidence of disease modification, this was accompanied by a decidedly mixed outcome on cognitive testing, with one aducanumab trial (EMERGE, NCT02484547) meeting its prespecified primary and secondary endpoints at the highest dose, whereas the other (ENGAGE, NCT02477800) did not achieve them. Although differences in cumulative dosing and uneven trial execution have been offered as explanations for this discrepancy, an FDA advisory committee was unconvinced and voted against approval. Nonetheless, the FDA granted an тАЬaccelerated approval,тАЭ citing robust amyloid lowering across both trials and an expectation that this should lead to less cognitive decline. It also required that a confirmatory trial be performed while marketing commences.
The controversy over aducanumab should be considered in the context of other recent AD immunotherapy trials. A large phase 2 trial of the monoclonal antibody lecanemab, designed to bind and clear A╬▓ protofibrils and oligomers, achieved its primary and secondary endpoints, including substantial amyloid plaque lowering and significantly less cognitive decline (10), and has advanced to phase 3 (NCT03887455). Another A╬▓ monoclonal antibody, gantenerumab, produced amyloid plaque reductions in phase 2 with less cognitive decline (11) and is in phase 3 (NCT03443973). Moreover, the antibody donanemab, which targets a low-abundance but aggregation-prone variant of A╬▓ with a modified amino terminus containing pyroglutamate-3, was recently shown in a moderate-sized phase 2 trial to markedly lower amyloid burden, accompanied by significant slowing of decline in psychometric tests and daily activities (12). Notably, donanemab conferred its cognitive effects in patients with relatively low tau burdens at trial entry (as judged by tau-PET), not in those with higher tau concentrations. Stratifying patients by tau burden was wise and could be used in future anti-A╬▓ trials. Unfortunately, however, both tau-PET and amyloid-PET only quantify fibrillar deposits, not soluble oligomers that appear to be responsible for neurotoxicity.
These four antibodies against A╬▓ unambiguously clear amyloid deposits from brain regions that are important for cognition, and this effect is accompanied by a variable 20 to 40% slowing of cognitive decline in 18-month trials. Collectively, these data represent the closest the AD field has come to a disease-modifying approach. So far, cognitive benefits are modest, and the challenge of assessing their clinical meaningfulness for patients and caregivers remains. But this challenge has been experienced in other chronic diseases, e.g., the controversy over the initial limited benefits of the antiretroviral drug zidovudine for HIV and AIDS when it was first approved (13).
Disease-modifying agents for AD are expected to slow cognitive decline more effectively the longerтАФand earlierтАФthey are given. Indeed, treating amyloid-positive individuals in the presymptomatic period is more likely to be efficacious. In mouse models of AD, early treatment with aducanumab reduced A╬▓ deposition and downstream neuropathology later in life. Gaining real-world experience with a first, albeit modest, treatment should encourage development of more potent second-generation agents.
Overnight, managing an untreatable, ultimately fatal disease has been converted into the complex challenge of offering treatment plans to myriad AD patients. Surprisingly, the indication on the aducanumab label initially read тАЬAlzheimer’s disease,тАЭ but after facing criticism, the FDA soon changed that to mild cognitive impairment and mild AD, mirroring the entry criteria for the phase 3 trials. AD clinicians will likely also require evidence of amyloid pathology. The latter can be established through amyloid-PET imaging, but this is not widely accessible, so CSF profiling will be relied upon to document the characteristic decrease in A╬▓42 monomers and increase in phospho-tau that has long been used to confirm AD.
A special challenge to clinicians will be considering amyloid-positive patients who are more impaired than those in the trials for treatment. AD practices offering aducanumab should establish transparent guidelines for patient eligibility, hopefully with limited variation among sites. The drug label specifies dose and infusion intervals, but criteria for how long to treat patients will evolve as any slowing of cognitive decline becomes apparent. The practical challenges of an infusible therapeutic will lead to subcutaneous formulations that can be administered at home. Parenthetically, the slower-release subcutaneous route may lessen the occurrence of the key adverse effect of antibodies against A╬▓: focal cerebral edema (ARIA-E), which is self-limited and asymptomatic in three-quarters of those who develop it and may be a sign of amyloid clearance or an inflammatory response at local vessels. Occasional microhemorrhages (ARIA-H) developed in a minority of those aducanumab recipients who had ARIA-E, and these appeared to be asymptomatic. The initial price of aducanumab (тИ╝$56,000/year) is very high and will need to be covered by insurance or national health care providers. These and other challenges in the march to implement the first approved AD therapeutic require thoughtful planning and resourcefulness, but this is just the process that patients and caregivers have long awaited.
A key advance has been the emergence of blood tests that can detect AD neuropathology. Plasma assays for certain fragments (14) and phospho-epitopes (15) of tau appear particularly promising, because tau alteration follows A╬▓ accumulation in those who develop AD symptoms. Comparison of various tau and A╬▓ plasma assays for their sensitivity in diagnosing AD and monitoring progression is needed. Accelerating the development of plasma biomarkers is critical to meet the challenge of screening innumerable patients globally for eligibility for AD-modifying agents.
Additional therapeutic approaches are crucial. Among small-molecule approaches, ╬▓-secretase inhibitors have been thwarted by mechanism-based side effects, although lower doses are being considered. An understudied class is the ╬│-secretase modulators that allosterically alter the conformation of presenilin and thereby shift APP processing from longer, amyloidogenic forms (A╬▓42, A╬▓43) to shorter, anti-amyloidogenic forms (A╬▓37, A╬▓38). Beyond A╬▓, effort is focused on slowing tau accumulation, e.g., by immunotherapy or antisense oligonucleotides. Modulating the pathological responses of macrophages and microglia is of great interest, given the strong genetic evidence for their involvement in AD. Nonpharmacological approaches toward preventing AD must also be pursued, including exercise, sleep hygiene, a Mediterranean diet, and intellectual and social enrichment.
For many chronic diseases, the initial therapeutic compounds have limited efficacy and are often steadily replaced by more effective drugs. The emerging immunotherapeutics slow the AD biological process but confer modest clinical benefit. The approval of aducanumab may provide a proof of concept that can be rapidly improved upon. It may also enable combination treatments, as is typical in chronic diseases. In therapeutics, as in life, one must walk before one can run.
Acknowledgments: D.J.S. is a director of and consultant for Prothena Biosciences.